1 2010-11-19 00:10:06 <Diablo-D3> ArtForz: http://pastebin.com/nHedqsT5
  2 2010-11-19 00:13:18 <gavinandresen> ArtForz: I think I'm running into the same problem with sends taking a long time that you ran into (more than 10 seconds to send).  My guess is SendCoins taking a very long time to decide what coins to use; I'll debug more tomorrow when I'm less tired.
  3 2010-11-19 00:13:54 <gavinandresen> (didn't/don't run into it with the Faucet because it doesn't send more than 0.50BTC at a time, only ran into it sending a few thousand BTC at once on testnet)
  4 2010-11-19 00:26:01 <Diablo-D3> art needs to write some sort of javascript calculator
  5 2010-11-19 00:26:20 <Diablo-D3> so I can just plug in seconds and blocks at diff 1 and get approx mhash
  6 2010-11-19 00:27:26 <Kiba> RESISTANCE IS FUTILE
  7 2010-11-19 00:29:37 <LobsterMan> so when diff increases do i get a new block to work on
  8 2010-11-19 00:29:38 <LobsterMan> ?
  9 2010-11-19 00:47:07 <xelister_> hmm anyone here would like to help me with either m0's or Diablos or other miner
 10 2010-11-19 00:47:54 <xelister_> to make it easly installable on both windows and linux(not just ubuntu)
 11 2010-11-19 00:48:20 <xelister_> and to auto-report and auto-send all generated to given address... and to show somewhere the miner's hashrate and number of successes
 12 2010-11-19 00:48:53 <xelister_> I could take m0m's in python, but python is not so trivially installable (problem to get pyopencl and such libs)
 13 2010-11-19 00:49:11 <xelister_> I was thinking of Diablos since java probably is easier to install,  but then it doesnt work on nvidias
 14 2010-11-19 00:49:42 <xelister_> Diablo-D3: is your mine working perfectly for radeons at least? i.e. no problems with some cards like 5970 etc? someone mentioned problems yesteday
 15 2010-11-19 01:13:44 <nanotube> ;;bc,stats
 16 2010-11-19 01:13:46 <gribble> Current Blocks: 92784 | Current Difficulty: 6866.89864897 | Next Difficulty At Block: 94752 | Next Difficulty In: 1968 blocks | Next Difficulty In About: 1 week, 5 days, 22 hours, 7 minutes, and 38 seconds | Next Difficulty Estimate: 7262.63832936
 17 2010-11-19 01:13:53 <nanotube> ooh, next diff is going down
 18 2010-11-19 01:16:04 <jgarzik> nanotube: ?
 19 2010-11-19 01:16:07 <Diablo-D3> xelister_: no, it works
 20 2010-11-19 01:16:27 <Diablo-D3> xelister_: but you have the use the build that isnt official yet
 21 2010-11-19 01:16:43 <Diablo-D3> WARGH WHAT THE FUCK IS WRONG WITH THIS PILE OF SHIT
 22 2010-11-19 01:16:51 <Diablo-D3> xelister_: oh, and mine works on nvidias now
 23 2010-11-19 01:23:39 <LobsterMan> Diablo-D3 it does?
 24 2010-11-19 01:23:53 <Diablo-D3> yes.
 25 2010-11-19 01:23:55 <LobsterMan> is the latest link available from your forum post? i want to give it a try
 26 2010-11-19 01:24:27 <Diablo-D3> no.
 27 2010-11-19 01:24:47 <LobsterMan> where can i find it?
 28 2010-11-19 01:24:55 <LobsterMan> or have you not released it yet?
 29 2010-11-19 01:25:03 <Diablo-D3> I haven't verified if it actually fucking works.
 30 2010-11-19 01:25:09 <LobsterMan> lol :\n48153
 31 2010-11-19 01:26:13 <Diablo-D3> its not the newest version I'm working with
 32 2010-11-19 01:26:36 <Diablo-D3> but it works on 2.1 and saturates 2x 5970 with no effort
 33 2010-11-19 01:26:50 <LobsterMan> i suppose i'll wait until you release something public <_<
 34 2010-11-19 01:26:56 <Diablo-D3> LobsterMan: oh, and, uh, THAT particular build may not totally function on anything but 5xxx
 35 2010-11-19 01:27:21 <Diablo-D3> LobsterMan: it does "work"
 36 2010-11-19 01:27:29 <Diablo-D3> but I think the G check code is fucked
 37 2010-11-19 01:27:30 <LobsterMan> lol
 38 2010-11-19 01:27:48 <Diablo-D3> so the actual mining shit works
 39 2010-11-19 01:36:11 <Diablo-D3> ...
 40 2010-11-19 01:36:16 <Diablo-D3> how the fuck does m0's miner even works
 41 2010-11-19 01:36:37 <LobsterMan> lol
 42 2010-11-19 01:36:41 <LobsterMan> what's interesting
 43 2010-11-19 01:36:48 <LobsterMan> ive been running his on testnet
 44 2010-11-19 01:36:56 <LobsterMan> and it seems that it generates 2 blocks
 45 2010-11-19 01:36:59 <Diablo-D3> it never does a write.
 46 2010-11-19 01:38:01 <LobsterMan> but it apparently does work
 47 2010-11-19 01:43:11 <LobsterMan> hm
 48 2010-11-19 01:43:15 <LobsterMan> maybe m0's doesnt work
 49 2010-11-19 01:43:38 <LobsterMan> one of the miners indicated a solution was found but bitcoin.exe didn't update with a new transaction
 50 2010-11-19 01:44:08 <LobsterMan> http://adterrasperaspera.com/images/DiabloMiner-test.zip ?
 51 2010-11-19 01:44:59 <Diablo-D3> yes
 52 2010-11-19 01:45:14 <yebyen> blarg!
 53 2010-11-19 01:45:46 <Diablo-D3> LobsterMan: if you're not on radeon 5xxx, edit the cl file in the jar at the obvious place at the top
 54 2010-11-19 01:46:02 <Diablo-D3> so, his code
 55 2010-11-19 01:46:05 <Diablo-D3> aalocates host ptr
 56 2010-11-19 01:46:12 <Diablo-D3> enques kernel
 57 2010-11-19 01:46:19 <Diablo-D3> enqueues read
 58 2010-11-19 01:46:33 <Diablo-D3> and never does clfinish
 59 2010-11-19 01:47:50 <ArtForz> he doesnt use async, and he waits on enqueue_read_buffer
 60 2010-11-19 01:47:52 <LobsterMan> Diablo-D3 change to:
 61 2010-11-19 01:47:54 <LobsterMan> ?
 62 2010-11-19 01:48:05 <Diablo-D3> LobsterMan: no comment out those two lines, uncomment the original
 63 2010-11-19 01:48:26 <LobsterMan> uncomment line1, then comment 2 and 3?
 64 2010-11-19 01:48:31 <Diablo-D3> LobsterMan: yes
 65 2010-11-19 01:48:35 <Diablo-D3> ArtForz: is wait magic or something?
 66 2010-11-19 01:49:15 <ArtForz> pyopencl returns a event for enqueue_whatever and kernel execution
 67 2010-11-19 01:49:29 <Diablo-D3> ahh
 68 2010-11-19 01:50:01 <Diablo-D3> I shouldnt need that if I set blocking to true
 69 2010-11-19 01:50:10 <ArtForz> blocking?
 70 2010-11-19 01:50:24 <Diablo-D3> third arg of enqueue read buffer
 71 2010-11-19 01:50:55 <ArtForz> yea, that should work, too
 72 2010-11-19 01:51:14 <ArtForz> but I dont think pyopencl supports that
 73 2010-11-19 01:51:18 <Diablo-D3> but he never does a write
 74 2010-11-19 01:51:46 <ArtForz> notice he doesn't pass a buffer as input
 75 2010-11-19 01:52:13 <ArtForz> he passes input in kernel args
 76 2010-11-19 01:52:35 <Diablo-D3> I pass the output buf as kernel args as well
 77 2010-11-19 01:52:50 <ArtForz> you apss a pointer to the buffer
 78 2010-11-19 01:53:01 <Diablo-D3> so does he
 79 2010-11-19 01:53:06 <ArtForz> check the kernel again
 80 2010-11-19 01:53:16 <Diablo-D3> output_buf is the last arg
 81 2010-11-19 01:53:19 <ArtForz> yes
 82 2010-11-19 01:53:23 <ArtForz> and he reads output buf
 83 2010-11-19 01:53:25 <Diablo-D3> output_buf = cl.Buffer(context, cl.mem_flags.WRITE_ONLY | cl.mem_flags.USE_HOST_PTR, hostbuf=output)
 84 2010-11-19 01:53:34 <Diablo-D3> which looks almost identical to what Im trying
 85 2010-11-19 01:53:37 <ArtForz> so?
 86 2010-11-19 01:54:10 <Diablo-D3> yeah, but how is he turning off H == 0?
 87 2010-11-19 01:54:20 <ArtForz> huH?
 88 2010-11-19 01:54:24 <Diablo-D3> merely doing buffer[0] == something that isnt 0 doesnt work.
 89 2010-11-19 01:54:36 <ArtForz> what? where?
 90 2010-11-19 01:54:43 <Diablo-D3> thats what Im saying
 91 2010-11-19 01:54:48 <Diablo-D3> where is he resetting the buffer after a hit
 92 2010-11-19 01:55:55 <ArtForz> now thats a good question
 93 2010-11-19 01:56:10 <Diablo-D3> if he doesnt do it, it'll repeatedly keep saying H == 0
 94 2010-11-19 01:56:17 <LobsterMan> this is weird, m0m's keeps getting hits on the testnet but it is NOT being indicated in my bitcoin.exe window
 95 2010-11-19 01:56:43 <LobsterMan> well, only some of them are
 96 2010-11-19 01:56:49 <LobsterMan> usually they appear right away
 97 2010-11-19 01:56:55 <LobsterMan> but it will get 2 hits, and only one of them shows up
 98 2010-11-19 01:57:35 <ArtForz> yeah, doesnt look like he resets output...
 99 2010-11-19 01:57:53 <ArtForz> wait
100 2010-11-19 01:58:02 <ArtForz> if output[0]: ... break
101 2010-11-19 01:58:18 <ArtForz> and the outer loop resets it
102 2010-11-19 01:58:46 <Diablo-D3> https://github.com/m0mchil/poclbm/blob/master/poclbm.py
103 2010-11-19 01:58:48 <Diablo-D3> what line?
104 2010-11-19 01:58:52 <ArtForz> 125
105 2010-11-19 01:59:02 <Diablo-D3> uhhh but it sets it to 0!
106 2010-11-19 01:59:11 <ArtForz> so?
107 2010-11-19 01:59:30 <Diablo-D3> oh.
108 2010-11-19 01:59:35 <Diablo-D3> hes just returning the nonce
109 2010-11-19 01:59:45 <ArtForz> yep
110 2010-11-19 01:59:52 <LobsterMan> here what do you make of this?
111 2010-11-19 02:00:34 <ArtForz> that looks... wrong
112 2010-11-19 02:00:38 <LobsterMan> yeah...
113 2010-11-19 02:00:44 <LobsterMan> i'm gonna post in in m0m's thread
114 2010-11-19 02:00:46 <LobsterMan> lol
115 2010-11-19 02:00:51 <ArtForz> I suspect getwork might have a problem with testnet
116 2010-11-19 02:00:57 <LobsterMan> hm
117 2010-11-19 02:01:44 <Diablo-D3> ArtForz: yeah but if I set output[0] to 1 (where I output 0 to indicate a hit)
118 2010-11-19 02:01:49 <xelister_> LobsterMan: m0's miner do work, I use it
119 2010-11-19 02:01:53 <Diablo-D3> it just gets forever stuck on that
120 2010-11-19 02:02:06 <LobsterMan> xelister_ did you see the screenshot i just posted?
121 2010-11-19 02:02:10 <ArtForz> Diablo-D3: err... wha?
122 2010-11-19 02:02:20 <LobsterMan> i've been running it for over a week on the real network and it's only generated a single block
123 2010-11-19 02:02:21 <Diablo-D3> ArtForz: I output two things, 0, and nonce
124 2010-11-19 02:02:30 <Diablo-D3> ArtForz: in the code, I check for output[0] == 0
125 2010-11-19 02:02:36 <Diablo-D3> if it is, I set it to 1
126 2010-11-19 02:02:42 <Diablo-D3> yet it never sees that I set it
127 2010-11-19 02:02:49 <Diablo-D3> and Im using a host_ptr'ed buffer like he is
128 2010-11-19 02:02:58 <ArtForz> *blank stare*
129 2010-11-19 02:03:10 <Diablo-D3> what?
130 2010-11-19 02:03:31 <ArtForz> so your kernel does if (H == 0 && output[0] == 0) {output[0] = 1; output[1] = nonce) or what?
131 2010-11-19 02:03:39 <Diablo-D3> no
132 2010-11-19 02:03:49 <xelister_> LobsterMan: wait 20 hours from miner hit to fully have usable coin
133 2010-11-19 02:04:04 <Diablo-D3> my kernel does if(H == 0) { output[0] = 0; output[1] = nonce; }
134 2010-11-19 02:04:17 <ArtForz> okay
135 2010-11-19 02:04:31 <LobsterMan> xelister_ what i'm saying is it's only generated one block for me so far, when according to the calculator and my hashrate i should have generated more than 1
136 2010-11-19 02:04:39 <LobsterMan> i suppose it's possible i am just very unlucky
137 2010-11-19 02:04:52 <xelister_> LobsterMan: single block per week sounds about right form medium nvidia
138 2010-11-19 02:05:11 <xelister_> whats the rate again? 30M ?
139 2010-11-19 02:05:37 <xelister_> 30M -> avg. time 11 days
140 2010-11-19 02:05:38 <Diablo-D3> ArtForz: so, in host code, I, like his, try output[0] = 1
141 2010-11-19 02:05:47 <xelister_> was 6 before today (diff inc)
142 2010-11-19 02:05:48 <LobsterMan> 50 per card, plus 4 from my cpu = 104mhash total
143 2010-11-19 02:06:12 <xelister_> well then you may be simply unlucky
144 2010-11-19 02:06:53 <Diablo-D3> argh!
145 2010-11-19 02:06:56 <Diablo-D3> fucking him
146 2010-11-19 02:06:58 <ArtForz> but he doesnt check == 1; he checks != 0
147 2010-11-19 02:07:00 <Diablo-D3> all he does is create a new buffer!
148 2010-11-19 02:07:09 <Diablo-D3> output[0] = base = 0
149 2010-11-19 02:07:14 <LobsterMan> Diablo-D3 what is it i need to change $@ to again? -user= and -pass= ??
150 2010-11-19 02:07:28 <Diablo-D3> LobsterMan: why are you changing that?
151 2010-11-19 02:07:31 <Diablo-D3> LobsterMan: just pass it to the script
152 2010-11-19 02:07:43 <xelister_> yeah I told to m0m's recently
153 2010-11-19 02:07:49 <xelister_> his code is unreadable ;)
154 2010-11-19 02:07:56 <LobsterMan> i thought i need to change the launch parameters for use in windows?
155 2010-11-19 02:08:07 <Diablo-D3> LobsterMan: no, you have to use whats in the op post of the thread
156 2010-11-19 02:08:10 <Diablo-D3> because windows sucks dick
157 2010-11-19 02:08:16 <xelister_> donkey's
158 2010-11-19 02:08:30 <xelister_> LobsterMan: want donkey vid? what price >_<
159 2010-11-19 02:08:38 <xelister_> that reminds me last time I was in a bar in mexico
160 2010-11-19 02:08:50 <ArtForz> fuck this, I'll never be able to make this fit in 6HE
161 2010-11-19 02:09:07 <Diablo-D3> 6he?
162 2010-11-19 02:09:17 <ArtForz> 6U
163 2010-11-19 02:09:32 <ArtForz> my rackmount 8*5970 design
164 2010-11-19 02:09:35 <Diablo-D3> ArtForz: so, question
165 2010-11-19 02:09:38 <Diablo-D3> ArtForz: you've seen my code right?
166 2010-11-19 02:09:57 <Diablo-D3> ArtForz: it does the full sha256(sha256) using java's own engine
167 2010-11-19 02:10:10 <Diablo-D3> should H == 0?
168 2010-11-19 02:10:12 <LobsterMan> Diablo-D3 it seems to be working now
169 2010-11-19 02:10:26 <Diablo-D3> LobsterMan: yes it is.
170 2010-11-19 02:10:32 <LobsterMan> how do i change the work size with yours?
171 2010-11-19 02:10:42 <LobsterMan> it's using 512 but i seem to do better with 256
172 2010-11-19 02:10:53 <Diablo-D3> LobsterMan: --help
173 2010-11-19 02:10:57 <ArtForz> so I guess 7U it is
174 2010-11-19 02:11:04 <xelister_> LobsterMan: you have working gpu miner on windows, that works with nv then?
175 2010-11-19 02:11:25 <Diablo-D3> long H = ((long)((0x000000FF & ((int)digestFirst.get(31))) << 24 |
176 2010-11-19 02:11:26 <LobsterMan> yes both m0mchil's and Diablo-D3's work for me
177 2010-11-19 02:11:33 <Diablo-D3> ArtForz: ^ shouldnt that output 0?
178 2010-11-19 02:11:53 <ArtForz> hrrrmm... not sure
179 2010-11-19 02:11:55 <ArtForz> wait a sec
180 2010-11-19 02:12:04 <Diablo-D3> ArtForz: I mean, if it does for the miner, shouldnt it for the host code?
181 2010-11-19 02:12:10 <xelister_> LobsterMan: can you describe how to get m0m's running on windows?
182 2010-11-19 02:12:15 <ArtForz> H is the LAST 4 bytes output by sha256
183 2010-11-19 02:12:24 <ArtForz> you flip it
184 2010-11-19 02:12:24 <Diablo-D3> that is the last 4.
185 2010-11-19 02:12:32 <xelister_> LobsterMan: assuming I will provide modified code of the miner - the .py file (other output format and so other things)
186 2010-11-19 02:12:42 <LobsterMan> xelister_ see first post on this page
187 2010-11-19 02:12:49 <ArtForz> doesnt that reverse the whole thnig?
188 2010-11-19 02:13:09 <Diablo-D3> ArtForz: thats bytes, and it endian flips the last int.
189 2010-11-19 02:13:21 <ArtForz> huH?
190 2010-11-19 02:13:23 <Diablo-D3> ArtForz: aaaand if H == 0, then the four bytes of H also all == 0
191 2010-11-19 02:13:33 <ArtForz> yea
192 2010-11-19 02:13:34 <Diablo-D3> ArtForz: java's sha256 routine outputs bytes not ints
193 2010-11-19 02:13:42 <brocktice> So, what'd I miss?
194 2010-11-19 02:13:48 <ArtForz> yes
195 2010-11-19 02:13:55 <ArtForz> so what does .flip() do?
196 2010-11-19 02:14:02 <Diablo-D3> oh, thats for buffer shit
197 2010-11-19 02:14:19 <Diablo-D3> flip resets position and marker to 0
198 2010-11-19 02:14:26 <Diablo-D3> and sets length to your former position
199 2010-11-19 02:14:41 <brocktice> I read back a little
200 2010-11-19 02:14:44 <brocktice> seems pretty heated
201 2010-11-19 02:15:04 <ArtForz> hrrrm
202 2010-11-19 02:15:31 <ArtForz> can't you just dump when whole thing when the kernel says H==0 and see which bytes if any actually *are* 0 ?
203 2010-11-19 02:15:50 <brocktice> ;;bc,estimate
204 2010-11-19 02:15:50 <Diablo-D3> ArtForz: why would I do that?
205 2010-11-19 02:15:50 <gribble> 7223.01518554
206 2010-11-19 02:15:53 <brocktice> hmm
207 2010-11-19 02:16:04 <LobsterMan> Diablo-D3 how long does it take to get the actual hashrate?
208 2010-11-19 02:16:14 <xelister_> brocktice: what that, next diff? how come "just" 7200
209 2010-11-19 02:16:15 <Diablo-D3> LobsterMan: anywhere from ten seconds to never.
210 2010-11-19 02:16:19 <brocktice> ;;bc,stats
211 2010-11-19 02:16:21 <gribble> Current Blocks: 92790 | Current Difficulty: 6866.89864897 | Next Difficulty At Block: 94752 | Next Difficulty In: 1962 blocks | Next Difficulty In About: 1 week, 5 days, 22 hours, 52 minutes, and 40 seconds | Next Difficulty Estimate: 7223.01518554
212 2010-11-19 02:16:21 <LobsterMan> heh
213 2010-11-19 02:16:30 <Diablo-D3> LobsterMan: its an average.
214 2010-11-19 02:16:32 <brocktice> I'm guessing people are getting discouraged from the new diff rate
215 2010-11-19 02:16:37 <LobsterMan> yeah i know
216 2010-11-19 02:16:41 <brocktice> s/rate/factor/
217 2010-11-19 02:16:50 <ArtForz> that and we're having a TX flood again
218 2010-11-19 02:16:52 <LobsterMan> mine is hanging at around 98.5mhash right now
219 2010-11-19 02:16:54 <brocktice> oh, that
220 2010-11-19 02:16:57 <brocktice> m0 fixed getwork though
221 2010-11-19 02:17:00 <LobsterMan> which is actually slightly slower than the total of m0m's
222 2010-11-19 02:17:03 <brocktice> but maybe people didn't update
223 2010-11-19 02:17:08 <ArtForz> = people still running old vers of getwork get near-0 hashrate
224 2010-11-19 02:17:11 <Diablo-D3> LobsterMan: m0's is inaccurate
225 2010-11-19 02:17:25 <LobsterMan> significantly inaccurate?
226 2010-11-19 02:17:26 <xelister_> brocktice: hm?  what he fixed?
227 2010-11-19 02:17:40 <Diablo-D3> LobsterMan: what does m0's say?
228 2010-11-19 02:17:48 <ArtForz> getwork rebuilt the block everytime a TX got added
229 2010-11-19 02:17:51 <brocktice> xelister_: it didn't deal well with TX flooding
230 2010-11-19 02:17:57 <ArtForz> stock limits it to once every 60 sec
231 2010-11-19 02:17:59 <LobsterMan> m0's averages from 47-51mhash per card
232 2010-11-19 02:18:00 <brocktice> now it only updates... yeah
233 2010-11-19 02:18:04 <LobsterMan> yours is giving me 98.5 for both
234 2010-11-19 02:18:18 <yebyen> ;blocks
235 2010-11-19 02:18:23 <yebyen> ;numblocks
236 2010-11-19 02:18:25 <brocktice> Diablo-D3: is yours making more than one block now?
237 2010-11-19 02:18:29 <brocktice> yebyen: bitbot is dead again
238 2010-11-19 02:18:29 <yebyen> does anyone know?
239 2010-11-19 02:18:31 <ArtForz> so if there's Txes trickling in constantly sec it rebuilt the block every fucking call
240 2010-11-19 02:18:31 <Diablo-D3> LobsterMan: 94 to 102
241 2010-11-19 02:18:39 <Diablo-D3> brocktice: Im still debugging that
242 2010-11-19 02:18:47 <Diablo-D3> LobsterMan: so mine says 98.5, so thats almost 102.
243 2010-11-19 02:18:48 <yebyen> i'm at 92164
244 2010-11-19 02:18:55 <ArtForz> and the whole thing is in a global lock, so miners were queueing up
245 2010-11-19 02:19:03 <LobsterMan> yeah Diablo-D3 it looks like they both give me about similar rates
246 2010-11-19 02:19:07 <brocktice> 92790
247 2010-11-19 02:19:14 <ArtForz> 92790
248 2010-11-19 02:19:24 <yebyen> ^_^ thumbs up
249 2010-11-19 02:19:26 <brocktice> yebyen: you can also talk to gribble
250 2010-11-19 02:19:30 <brocktice> ;;bc,stats
251 2010-11-19 02:19:32 <gribble> Current Blocks: 92790 | Current Difficulty: 6866.89864897 | Next Difficulty At Block: 94752 | Next Difficulty In: 1962 blocks | Next Difficulty In About: 1 week, 5 days, 22 hours, 52 minutes, and 40 seconds | Next Difficulty Estimate: 7223.01518554
252 2010-11-19 02:19:45 <yebyen> brocktice: thanks :)
253 2010-11-19 02:20:06 <brocktice> np, thank nanotube
254 2010-11-19 02:20:19 <LobsterMan> Diablo-D3 i'm gonna test yours on testnet now and see what happens ;]
255 2010-11-19 02:20:26 <xelister_> ArtForz: what are the symptoms of using old get-work without that correct Tx thing?
256 2010-11-19 02:20:34 <Diablo-D3> LobsterMan: it'll run, but not make any valid attempts
257 2010-11-19 02:20:44 <Diablo-D3> Attempt 92 27 -125 22 -114 4 32 -14 125 -75 29 50 85 36 104 -115 35 6 45 116 126 12 -5 -18 -16 43 -12 -19 -63 22 -28 -14
258 2010-11-19 02:20:46 <Diablo-D3> ArtForz: ^
259 2010-11-19 02:20:47 <LobsterMan> so it won't generate any blocks at all on testnet?
260 2010-11-19 02:20:51 <ArtForz> wel??l, that and near 100% of one core used
261 2010-11-19 02:20:53 <Diablo-D3> ArtForz: none of that looks like an H == 0
262 2010-11-19 02:20:58 <ArtForz> nope
263 2010-11-19 02:21:02 <Diablo-D3> LobsterMan: it wont generate on any net
264 2010-11-19 02:21:08 <Diablo-D3> LobsterMan: trying to debug that
265 2010-11-19 02:21:12 <LobsterMan> oh :\n48402
266 2010-11-19 02:21:22 <ArtForz> diablo: hrrrm.. are you properly flipping the nonce for input to your java sha?
267 2010-11-19 02:21:24 <xelister_> your miner does not generate anything on any net?
268 2010-11-19 02:21:25 <xelister_> good job ;)
269 2010-11-19 02:21:34 <brocktice> yebyen: ;;botsnack
270 2010-11-19 02:21:36 <Diablo-D3> ArtForz: shouldn't have to
271 2010-11-19 02:21:40 <brocktice> ;;botsnack
272 2010-11-19 02:21:40 <gribble> Why not send me some bitcoins instead, to this address: 1MgD6rah5zUgEGYZnNmdpnXMaDR3itKYzU :)
273 2010-11-19 02:21:42 <Diablo-D3> ArtForz: I dont flip it feeding it to opencl
274 2010-11-19 02:21:47 <LobsterMan> well Diablo-D3 at least you know it works on windows now :)
275 2010-11-19 02:21:52 <gribble> Why not send me some bitcoins instead, to this address: 1MgD6rah5zUgEGYZnNmdpnXMaDR3itKYzU :)
276 2010-11-19 02:21:52 <yebyen> ;;botsnack
277 2010-11-19 02:21:54 <yebyen> hey that's opportunistic
278 2010-11-19 02:21:55 <xelister_> gribble: instead of what?
279 2010-11-19 02:21:56 <ArtForz> but isnt opencl LE?
280 2010-11-19 02:22:02 <Diablo-D3> LobsterMan: not only that, it works on nvidia and stream sdk 2.1
281 2010-11-19 02:22:03 <yebyen> how about a drink instead
282 2010-11-19 02:22:06 <brocktice> nanotube puts in the effort to run it
283 2010-11-19 02:22:13 <brocktice> It's not my address :)
284 2010-11-19 02:22:20 <xelister_> gribble: are you 18 girl with a webcam
285 2010-11-19 02:22:32 <Diablo-D3> ArtForz: no, its BE, with automatic flipping for LE host/client
286 2010-11-19 02:22:40 <ArtForz> okay
287 2010-11-19 02:22:43 <Diablo-D3> ArtForz: and Java runs this sha256 shit in BE
288 2010-11-19 02:22:44 <Diablo-D3> magically
289 2010-11-19 02:22:50 <ArtForz> well, sha256 is BE
290 2010-11-19 02:22:58 <ArtForz> well, kinda
291 2010-11-19 02:22:59 <Diablo-D3> sha256 is whatever the fuck the person coded it to be
292 2010-11-19 02:23:09 <ArtForz> it's u32s all throughout
293 2010-11-19 02:23:14 <ArtForz> but in/out is defined to be BE
294 2010-11-19 02:23:18 <Diablo-D3> yeah
295 2010-11-19 02:23:19 <Diablo-D3> but like
296 2010-11-19 02:23:21 <Diablo-D3> you have getwork
297 2010-11-19 02:23:28 <Diablo-D3> which outputs an invalid sha256 state
298 2010-11-19 02:23:30 <Diablo-D3> because its LE
299 2010-11-19 02:23:49 <Diablo-D3> which I fixup
300 2010-11-19 02:23:58 <ArtForz> well, with a LE host the LE state makes perfect sense
301 2010-11-19 02:24:03 <Diablo-D3> yup
302 2010-11-19 02:24:09 <Diablo-D3> but I have to swap it to bE
303 2010-11-19 02:24:12 <ArtForz> yep
304 2010-11-19 02:24:12 <Diablo-D3> well no
305 2010-11-19 02:24:16 <Diablo-D3> I have to swap it FROM be
306 2010-11-19 02:24:22 <ArtForz> isn't it fun?
307 2010-11-19 02:24:26 <Diablo-D3> since java's parse shit ONLY reads be
308 2010-11-19 02:24:30 <Diablo-D3> no matter what your host is
309 2010-11-19 02:24:36 <ArtForz> yeah
310 2010-11-19 02:24:41 <Diablo-D3> because it assumes text is always BE since thats the way SHIT IS SUPPOSED TO WORK
311 2010-11-19 02:24:57 <ArtForz> node sends it LE, java decodes it as BE, so you have to byteswap it
312 2010-11-19 02:25:06 <Diablo-D3> yup
313 2010-11-19 02:25:15 <Diablo-D3> and Im fucked if anyone gets BE hosts
314 2010-11-19 02:25:18 <brocktice> Diablo-D3: do you need a drink man?
315 2010-11-19 02:25:25 <brocktice> Or, like, a blunt?
316 2010-11-19 02:25:30 <Diablo-D3> this app is 100% endian fuck you.
317 2010-11-19 02:25:41 <Diablo-D3> its not merely unsafe, it'll rape your female relatives.
318 2010-11-19 02:25:43 <xelister_> lets make a bothendian app
319 2010-11-19 02:25:53 <brocktice> xelister_: middleendian?
320 2010-11-19 02:25:59 <xelister_> no, it would use HH notation
321 2010-11-19 02:26:05 <brocktice> msb is in the midle
322 2010-11-19 02:26:07 <ArtForz> NUXI endian ?
323 2010-11-19 02:26:09 <brocktice> then it alternates sides after that
324 2010-11-19 02:26:17 <ArtForz> wasnt that PDP/11?
325 2010-11-19 02:26:20 <brocktice> lol
326 2010-11-19 02:26:34 <brocktice> I've read papers that say the've used PDP/11s
327 2010-11-19 02:26:35 <Diablo-D3> middle endian: swap the bytes AND the bits
328 2010-11-19 02:26:37 <brocktice> and I've seen them
329 2010-11-19 02:26:43 <ArtForz> Diable-D3: nope
330 2010-11-19 02:26:45 <brocktice> but I've never actually seen one plugged in and operating
331 2010-11-19 02:26:53 <Diablo-D3> brocktice: neither has anyone else
332 2010-11-19 02:26:54 <Diablo-D3> dohohohoho
333 2010-11-19 02:27:08 <brocktice> Not anyone from my generation, probably
334 2010-11-19 02:27:15 <ArtForz> BE: 4321, LE: 1234, ME: 3412
335 2010-11-19 02:27:19 <brocktice> unless some poor coder got hired for an arcane maintenance job
336 2010-11-19 02:27:30 <ArtForz> or 2143, but I think no one actually did that
337 2010-11-19 02:27:30 <xelister_> bothendian: 3434
338 2010-11-19 02:27:32 <Diablo-D3> ArtForz: eww
339 2010-11-19 02:27:36 <brocktice> "Here, this PDP/11 is mission-critical, keep it working"
340 2010-11-19 02:27:42 <xelister_> bothendian is what real people use
341 2010-11-19 02:28:08 <brocktice> "what, your graphing calculator could replace it?"
342 2010-11-19 02:28:13 <ArtForz> what network protocol was bothendian again?
343 2010-11-19 02:28:38 <ArtForz> 1234 was sent as 12344321
344 2010-11-19 02:28:59 <xelister_> now, I just made it up
345 2010-11-19 02:29:18 <xelister_> real man use 3434 and they dont care about 1 and 2 octets =)
346 2010-11-19 02:29:49 <ArtForz> I prefer 4-banger endian, 1-3-4-2
347 2010-11-19 02:33:30 <Diablo-D3> ArtForz: I suspect I figured out the problem
348 2010-11-19 02:33:56 <Diablo-D3> ArtForz: I'm not flipping the whole block when feeding it to the java engine
349 2010-11-19 02:34:01 <Diablo-D3> since it expects big endian
350 2010-11-19 02:36:02 <ArtForz> just byteflip dwords until it works
351 2010-11-19 02:36:09 <Diablo-D3> yup
352 2010-11-19 02:36:36 <xelister_> this is one problem with testing bitcoin applications
353 2010-11-19 02:36:53 <xelister_> it takes forever to get block if testing small miners
354 2010-11-19 02:36:55 <xelister_> on real net
355 2010-11-19 02:37:18 <ArtForz> esue esttent
356 2010-11-19 02:37:25 <ArtForz> err
357 2010-11-19 02:37:29 <ArtForz> use -testnet
358 2010-11-19 02:37:38 <Diablo-D3> fail
359 2010-11-19 02:37:49 <xelister_> its not fully the same to test real application
360 2010-11-19 02:38:01 <Diablo-D3> xelister_: yes it is
361 2010-11-19 02:38:02 <xelister_> other port  number,  someone said getwork may ahve problem with that and so on and so on
362 2010-11-19 02:38:03 <ArtForz> it's same enough for me
363 2010-11-19 02:38:22 <xelister_> hm but yeah I suppose this is not that bad...
364 2010-11-19 02:38:36 <xelister_> so there is one testnet? and its diff also rises?
365 2010-11-19 02:38:40 <xelister_> but slower?
366 2010-11-19 02:38:54 <LobsterMan> Diablo-D3 one thing i noticed about yours.....yours seemed to run a bit cooler on my cards than m0m's does
367 2010-11-19 02:39:05 <Diablo-D3> superior card is superior
368 2010-11-19 02:39:06 <Diablo-D3> er
369 2010-11-19 02:39:09 <Diablo-D3> superior miner
370 2010-11-19 02:39:14 <LobsterMan> m0m's makes me max out around 83-84???, but yours was only like 80
371 2010-11-19 02:39:19 <ArtForz> testnet diff started lower and works the same way main diff does
372 2010-11-19 02:39:33 <xelister_> so in some time also on testnet it will be hard
373 2010-11-19 02:39:39 <ArtForz> no
374 2010-11-19 02:39:46 <ArtForz> because nearly no one is really generating on testnet
375 2010-11-19 02:39:51 <ArtForz> it's at 5.33 as I threw a 5770 at it for a bit to get diff above 1.0
376 2010-11-19 02:40:03 <Diablo-D3> Attempt 1 0 0 0 -7 28 30 110 25 -126 111 -48 -17 -32 52 24 120 52 -113 82 54 -26 -101 -128 0 29 109 62 86 83 -53 45
377 2010-11-19 02:40:04 <Diablo-D3> hrm
378 2010-11-19 02:40:20 <Diablo-D3> ttempt 1 0 0 0 -7 28 30 110 25 -126 111 -48 -17 -32 52 24 120 52 -113 82 54 -26 -101 -128 0 29 109 62 86 83 -53 45
379 2010-11-19 02:40:26 <ArtForz> 0 0 0 -7 ? looks nearly right
380 2010-11-19 02:40:30 <Diablo-D3> wait, what?
381 2010-11-19 02:40:35 <Diablo-D3> the same attempt twice?
382 2010-11-19 02:40:45 <ArtForz> l??l
383 2010-11-19 02:41:41 <Diablo-D3> oh duh
384 2010-11-19 02:41:47 <Diablo-D3> Im not printing the right thing now
385 2010-11-19 02:43:29 <LobsterMan> l??l
386 2010-11-19 02:43:30 <LobsterMan> lol
387 2010-11-19 02:43:31 <LobsterMan> :P
388 2010-11-19 02:45:20 <LobsterMan> he's said it doesn't work yet
389 2010-11-19 02:45:21 <LobsterMan> :P
390 2010-11-19 02:46:46 <xelister_> is this cause that you got so far 1 block after a week ?
391 2010-11-19 02:47:39 <xelister_> hm  miner seems to hang flash videos
392 2010-11-19 02:47:52 <Diablo-D3> xelister_: flash sucks
393 2010-11-19 02:47:54 <ArtForz> sash flucks
394 2010-11-19 02:47:59 <xelister_> actually, this may be considered advantage :)  if not only the stupid firefox hangs with it
395 2010-11-19 02:48:02 <Diablo-D3> lol
396 2010-11-19 02:48:08 <xelister_> yes, we all know, flash is shit
397 2010-11-19 02:48:11 <Diablo-D3> xelister_: erm, use a newer firefox
398 2010-11-19 02:48:27 <Diablo-D3> xelister_: firefox 3.6.x has out of process plugin execution now
399 2010-11-19 02:48:36 <xelister_> Diablo-D3: thats comming from a Debian guy? ;) Im using the one from 10.10 ubuntu
400 2010-11-19 02:49:00 <Diablo-D3> ArtForz: Im on drugs, how many bytes is a sha256 output? 32?
401 2010-11-19 02:49:10 <ArtForz> ya
402 2010-11-19 02:49:47 <xelister_> there should be medical-gajna-2-bitcoin trader
403 2010-11-19 02:52:38 <Kiba> medical-nin?
404 2010-11-19 02:52:53 <Kiba> strange combo
405 2010-11-19 02:53:24 <Diablo-D3> ArtForz: this is entirely fucked
406 2010-11-19 02:53:55 <Diablo-D3> ArtForz: how many bytes is the block header? 128?
407 2010-11-19 03:04:16 <nanotube> jgarzik: i meant it went down from earlier today. the estimate was as high as 13k earlier, a few blocks after the diff went up to 6k
408 2010-11-19 03:05:07 <jgarzik> ahh
409 2010-11-19 03:26:00 <LobsterMan> diff is higher now than what was estimated a few days back
410 2010-11-19 03:26:06 <LobsterMan> it was estimated to be like 6400
411 2010-11-19 03:58:00 <Diablo-D3> AHAAA
412 2010-11-19 03:58:01 <Diablo-D3> AAHh
413 2010-11-19 03:58:03 <Diablo-D3> MOTHERFUCKING
414 2010-11-19 03:58:04 <Diablo-D3> HAAA
415 2010-11-19 03:58:10 <Diablo-D3> -61 -50 123 -69 19 -43 51 34 91 -106 9 33 -42 -97 -8 125 -108 -122 -123 -61 64 99 17 91 91 -41 1 114 0 0 0 0
416 2010-11-19 03:59:53 <Diablo-D3> I figured out the bug
417 2010-11-19 04:00:01 <Diablo-D3> I was feeding it the whole block header
418 2010-11-19 04:00:09 <Diablo-D3> all 128 bytes of the 80 byte header
419 2010-11-19 04:43:26 <ByteCoin> theymos: You at keyboard?
420 2010-11-19 04:43:33 <ByteCoin> Calling theymos!!!
421 2010-11-19 04:43:50 <ByteCoin> How do you make it beep?
422 2010-11-19 04:43:59 <ByteCoin> :call theymos
423 2010-11-19 04:44:16 <ByteCoin> ::help
424 2010-11-19 04:44:19 <ByteCoin> :help
425 2010-11-19 04:44:29 <ByteCoin> IRC not my strong poiny
426 2010-11-19 04:47:08 <nanotube> ByteCoin: haha you don't make it beep.
427 2010-11-19 04:47:24 <nameless> |nanotube: sure you do
428 2010-11-19 04:47:35 <nameless> |You just have to know how
429 2010-11-19 04:47:38 <nanotube> nameless|: not if he doesn't have the client set to beep :P
430 2010-11-19 04:47:48 <ByteCoin> I think it beeps if you send a private message.
431 2010-11-19 04:47:50 <nanotube> nameless|: enlighten me, in that case. please. :)
432 2010-11-19 04:48:11 <nanotube> nameless|: (note that i use xchat, and i have turned off all visual notifications to nick highlights :) )
433 2010-11-19 04:48:13 <ByteCoin> It doesn't matter now. I've mailed him instead
434 2010-11-19 04:48:31 <ByteCoin> nameless: Make me beep
435 2010-11-19 04:49:01 <ByteCoin> I'm not getting it.....
436 2010-11-19 04:49:10 <ByteCoin> Do it!
437 2010-11-19 04:49:14 <ByteCoin> Do it now!
438 2010-11-19 04:49:20 <nanotube> haha
439 2010-11-19 04:49:46 <ByteCoin> *sulks*
440 2010-11-19 04:49:50 <nameless> |I love it when people bel flood and everyone gets pissed off.
441 2010-11-19 04:50:05 <nameless> |Me, I have irssi and screen set up to ignore bel
442 2010-11-19 04:50:50 <ne0futur> same here
443 2010-11-19 04:51:54 <nameless> |on that note
444 2010-11-19 04:52:02 <nameless> |I have trolled channels with bel
445 2010-11-19 04:52:12 <nameless> |But I was trolling them with lots of stuff before
446 2010-11-19 04:53:45 <nanotube> so there are irc clients that actually beep on ^g?
447 2010-11-19 04:54:07 <nameless> |irssi does if you configure it too
448 2010-11-19 04:54:10 <nameless> |and yes, there are
449 2010-11-19 04:54:19 <nameless> |lots of them
450 2010-11-19 04:54:36 <nanotube> hmm
451 2010-11-19 04:54:39 <nameless> |On one network I'm on, the server kicks you if you use bel
452 2010-11-19 04:54:43 <nanotube> hehe
453 2010-11-19 04:54:53 <nameless> |rather, a bot spawned by the server
454 2010-11-19 04:54:59 <nameless> |but the bot is U:Lined
455 2010-11-19 05:04:41 <ByteCoin> Hi freemoney.
456 2010-11-19 05:04:49 <FreeMoney> hi
457 2010-11-19 05:04:50 <ByteCoin> I;ve just replied to your porst
458 2010-11-19 05:05:08 <ByteCoin> Bt you can tell me who the flooder is here if you want
459 2010-11-19 05:05:20 <ByteCoin> How did they get you to give them coins?
460 2010-11-19 05:05:28 <ByteCoin> Did they have a pretext
461 2010-11-19 05:05:31 <ByteCoin> ?
462 2010-11-19 05:05:34 <FreeMoney> MrBurns
463 2010-11-19 05:05:53 <FreeMoney> he just said he wanted to see how it worked and how long it took etc
464 2010-11-19 05:06:05 <ByteCoin> Is that his name on the forum?
465 2010-11-19 05:06:10 <FreeMoney> and that he couldn't use the faucet because he was on TOR
466 2010-11-19 05:06:12 <FreeMoney> yes
467 2010-11-19 05:06:23 <FreeMoney> he has only the one post
468 2010-11-19 05:06:45 <ByteCoin> I'm just trying to look it up...
469 2010-11-19 05:07:20 <FreeMoney> in bitcoin discussion
470 2010-11-19 05:07:27 <FreeMoney> "new to bitcoins....
471 2010-11-19 05:07:35 <ByteCoin> Got it thx
472 2010-11-19 05:08:37 <Kiba> new forum section..for all the bounty that has been poping up..
473 2010-11-19 05:10:39 <ByteCoin> I wonder what the fallout of the flooding will be....
474 2010-11-19 05:10:52 <ByteCoin> It doesn't seem to be stopping!
475 2010-11-19 05:11:40 <ByteCoin> Also, it seems likely to me that MrBurns is an alias of someone fairly technically adept probably already posting on the forum.
476 2010-11-19 05:15:40 <Diablo-D3> Yesssss
477 2010-11-19 05:16:16 <ByteCoin> Did you just get a block Diablo?
478 2010-11-19 05:16:22 <OneFixt> ;;bc,stats
479 2010-11-19 05:16:25 <gribble> Current Blocks: 92816 | Current Difficulty: 6866.89864897 | Next Difficulty At Block: 94752 | Next Difficulty In: 1936 blocks | Next Difficulty In About: 1 week, 4 days, 14 hours, 33 minutes, and 44 seconds | Next Difficulty Estimate: 7954.12623257
480 2010-11-19 05:21:01 <Diablo-D3> ByteCoin: no
481 2010-11-19 05:21:22 <ByteCoin> Someone's generating quite fast.....
482 2010-11-19 05:22:05 <nanotube> probably ArtForz :)
483 2010-11-19 05:22:19 <ByteCoin> about 1 minute since the last block.. I know about the probabiltiies yes it could just be chance
484 2010-11-19 05:22:19 <nanotube> well... the whole network overall, really. :)
485 2010-11-19 05:22:41 <ByteCoin> the next difficulty seems to indicate that there's not been a step change in processing power
486 2010-11-19 05:28:35 <Diablo-D3> ByteCoin: wait until it gets there first
487 2010-11-19 05:29:12 <ByteCoin> Perhaps I should have said - not been a large step change
488 2010-11-19 05:29:36 <Diablo-D3> yeah, but it can change in the next week
489 2010-11-19 05:29:43 <Diablo-D3> should be closer to over 9000
490 2010-11-19 05:30:15 <nanotube> heh yea... someone buys 4 5970s, and boom, a bunch of extra ghps
491 2010-11-19 05:30:21 <Diablo-D3> or 10
492 2010-11-19 05:32:36 <gribble> {"ticker":{"high":0.283,"low":0.2362,"vol":37480,"buy":0.2683,"sell":0.28,"last":0.28}}
493 2010-11-19 05:32:36 <nanotube> ;;bc,mtgox
494 2010-11-19 05:44:45 <ByteCoin> I can hardly believe that one of the recent blocks was generated four seconds!! after it's predecessor!
495 2010-11-19 05:45:05 <ByteCoin> Even allowing for some clock skew
496 2010-11-19 05:45:42 <nanotube> hm yea that does seem awful fast
497 2010-11-19 05:45:45 <lfm> the net is quite efficient at distributing those blocks. you shouldnt be surprized
498 2010-11-19 05:46:28 <nanotube> it's just 'improbable' :) for any given block, that is. over the whole block chain, probably not so much. :)
499 2010-11-19 05:47:21 <ByteCoin> I'm not looking at the whole block chain! Another recent one was generated 20 seconds after the previous one!
500 2010-11-19 05:47:51 <OneFixt> ;;bc,stats
501 2010-11-19 05:47:53 <gribble> Current Blocks: 92822 | Current Difficulty: 6866.89864897 | Next Difficulty At Block: 94752 | Next Difficulty In: 1930 blocks | Next Difficulty In About: 1 week, 4 days, 8 hours, 31 minutes, and 45 seconds | Next Difficulty Estimate: 8105.00858039
502 2010-11-19 05:48:08 <nanotube> mm, the estimate is growing
503 2010-11-19 05:51:19 <lfm> so 10k difficulty in 4 weeks
504 2010-11-19 05:51:27 <lfm> ?
505 2010-11-19 05:52:29 <nanotube> mm, well, the difficulty estimate has actually been growing... so it may get to ~10k even within this block chunk.
506 2010-11-19 06:47:52 <OneFixt> ;;bc,stats
507 2010-11-19 06:47:54 <gribble> Current Blocks: 92831 | Current Difficulty: 6866.89864897 | Next Difficulty At Block: 94752 | Next Difficulty In: 1921 blocks | Next Difficulty In About: 1 week, 4 days, 1 hour, 25 minutes, and 37 seconds | Next Difficulty Estimate: 8283.07426245
508 2010-11-19 08:28:07 <Akiraa> Is there a GPU bitcoin binary available?
509 2010-11-19 08:36:20 <LobsterMan> several
510 2010-11-19 08:36:27 <LobsterMan> check the forums
511 2010-11-19 11:25:11 <anarchyx> ;estimate
512 2010-11-19 11:25:17 <anarchyx> ;;estimate
513 2010-11-19 11:25:17 <gribble> Error: "estimate" is not a valid command.
514 2010-11-19 11:25:33 <anarchyx> ;;bc,stats
515 2010-11-19 11:25:39 <gribble> Current Blocks: 92858 | Current Difficulty: 6866.89864897 | Next Difficulty At Block: 94752 | Next Difficulty In: 1894 blocks | Next Difficulty In About: 1 week, 4 days, 11 hours, 47 minutes, and 36 seconds | Next Difficulty Estimate: 7859.69044161
516 2010-11-19 11:25:43 <anarchyx> ;;help
517 2010-11-19 11:25:43 <gribble> The bot responds when you start a line with the ! character. A good starting point for exploring the bot is the !facts command. You can also visit the bot's website for a list of help topics and documentation: http://gribble.sourceforge.net/
518 2010-11-19 11:25:59 <anarchyx> ;;facts
519 2010-11-19 11:25:59 <gribble> To see a nice sortable web view of all factoids, click here: http://gribble.dreamhosters.com/viewfactoids.php?db=%23bitcoin-dev || To see a list of the most popular factoids, run !rank || To search factoids, run !factoids search <yoursearchterm>
520 2010-11-19 11:42:06 <nanotube> anarchyx: yea there are no factoids for this channel... :)
521 2010-11-19 12:21:34 <xelister_> LobsterMan: on which forum thread you based your installation?  is all there actuall, or did something needs work arounds
522 2010-11-19 12:49:19 <joe_1> artforz please turn on some more radeons, i need confirmations fast
523 2010-11-19 13:14:22 <xelister_> joe_1: ;)
524 2010-11-19 13:26:17 <brocktice> the difficulty estimate is oscillating
525 2010-11-19 13:26:26 <brocktice> There's too much noise
526 2010-11-19 13:26:32 <brocktice> We don't have a long history at this difficulty
527 2010-11-19 13:26:54 <brocktice> Quick PSU question, if a PSU is rated for '1kW', is that internally, or at the plug?
528 2010-11-19 13:34:13 <brocktice> looks like it's internally
529 2010-11-19 13:34:17 <brocktice> with caveats
530 2010-11-19 13:43:06 <anarchyx> ;;bc,stats
531 2010-11-19 13:43:08 <gribble> Current Blocks: 92874 | Current Difficulty: 6866.89864897 | Next Difficulty At Block: 94752 | Next Difficulty In: 1878 blocks | Next Difficulty In About: 1 week, 4 days, 6 hours, 30 minutes, and 44 seconds | Next Difficulty Estimate: 7945.43968873
532 2010-11-19 13:43:58 <anarchyx> i hope the difficulty will start levelling off
533 2010-11-19 13:44:02 <anarchyx> its getting crazy
534 2010-11-19 13:44:20 <brocktice> It must be
535 2010-11-19 13:44:29 <brocktice> because we're at about 7-8 blocks per hour
536 2010-11-19 13:44:32 <brocktice> instead of 10 or 11
537 2010-11-19 13:56:34 <Kiba> so
538 2010-11-19 13:56:48 <Kiba> does anybody have any experience with mybitcoin SGI merchant API or whatever?
539 2010-11-19 14:07:11 <brocktice> woohoo, after bumping core voltage back up to 1.175 I'm at 1450 mhash/s
540 2010-11-19 14:07:24 <brocktice> maybe if Diablo's fixed his miner problems I can even exced that.
541 2010-11-19 15:18:52 <ByteCoin> Hi gavin
542 2010-11-19 15:21:58 <ByteCoin> It looks like the flood has taken a turn for the worse
543 2010-11-19 15:22:38 <ByteCoin> As of 92879 the blocks all have about 220 transactions in them.
544 2010-11-19 15:23:53 <ByteCoin> If the transactions are 227 bytes long this corresponds with filling up the 50k fee free portion of the block...
545 2010-11-19 15:24:28 <m0mchil> they are low priority
546 2010-11-19 15:25:04 <m0mchil> but anyway, still bad... too much disk space
547 2010-11-19 15:28:17 <m0mchil> spam fee, proportional to the lowest officially spendable ammount would solve this
548 2010-11-19 15:28:45 <edcba> what do you mean ?
549 2010-11-19 15:30:12 <brocktice> edcba: He means making it cost a few bitcoins to TX spam
550 2010-11-19 15:30:24 <edcba> how do you define spam ?
551 2010-11-19 15:30:26 <m0mchil> say, for now officially lowest spendable amount is 0.01... placing mandatory 0.001 for example will solve the problem with countless meaningless transactions filling the block chain
552 2010-11-19 15:30:55 <edcba> ok you want a tx fee being fraction of a COIN
553 2010-11-19 15:31:02 <jgarzik> m0mchil: FWIW satoshi said he wanted some tranche of transactions to be free
554 2010-11-19 15:31:13 <jgarzik> m0mchil: you know I agree with you, of course :)
555 2010-11-19 15:32:04 <m0mchil> yup... this can only get worse...
556 2010-11-19 15:33:04 <edcba> i don't really like having tx fees like that
557 2010-11-19 15:33:15 <edcba> i would do it some other way
558 2010-11-19 15:33:32 <m0mchil> thats why we are discussing it
559 2010-11-19 15:34:00 <edcba> just reserve like some 1000 TX buffer
560 2010-11-19 15:34:12 <edcba> TX with most fees gets in
561 2010-11-19 15:34:20 <edcba> others are discarded
562 2010-11-19 15:39:24 <edcba> then tx spam will remain local to neighbour spammer when buffer filled
563 2010-11-19 15:39:45 <edcba> or he will have to  pay fee
564 2010-11-19 15:39:49 <jgarzik> free TX sits around until some miner who does not require fees builds a block including that TX
565 2010-11-19 15:40:04 <jgarzik> each block only permits certain amount of free TX
566 2010-11-19 15:40:30 <ArtForz> isnt that what we already have?
567 2010-11-19 15:40:38 <jgarzik> yes, just stating the current situation
568 2010-11-19 15:41:43 <brocktice> ArtForz: My PSU is rated 1kW, now drawing about 1.005 kW at the plug, should be OK because rating is for output, not input, right?
569 2010-11-19 15:41:52 <jgarzik> in theory, spam transactions linger around, wasting everybody's resources, until they reach a block
570 2010-11-19 15:42:12 <ArtForz> brocktice: yeah, but make sure your PSU has plenty of airflow
571 2010-11-19 15:42:13 <jgarzik> but maybe too-much spam implies spam TXs will disappear due to age, if they are so large in number?
572 2010-11-19 15:42:29 <m0mchil> see this '0.01 BTC fee if sending any transaction less than 0.01 BTC. This is to help prevent DoS attacks against the network.'
573 2010-11-19 15:42:39 <brocktice> ArtForz: It gets a decent amount, but I worry that the incoming air (out of the radiator) is not cool enough to start with :(
574 2010-11-19 15:43:01 <edcba> m0mchil: definitely flawed :)
575 2010-11-19 15:43:13 <jgarzik> m0mchil: that prevents small-transaction spam, but clearly does not prevent me sending 0.01 BTC to myself over and over again
576 2010-11-19 15:43:16 <brocktice> Can't you flood the network with transactions to yourself?
577 2010-11-19 15:43:17 <brocktice> yeah
578 2010-11-19 15:43:19 <jgarzik> yes
579 2010-11-19 15:43:33 <m0mchil> not taking in account sending to yourself
580 2010-11-19 15:43:57 <ArtForz> stupid idea: min fee-free value rises with block size
581 2010-11-19 15:44:02 <m0mchil> and because there is no way to detect someone sending to himself - I can't see any other solution
582 2010-11-19 15:44:34 <brocktice> m0mchil: the client detects it somehow
583 2010-11-19 15:44:41 <brocktice> although I guess one could just rewrite it
584 2010-11-19 15:44:55 <jgarzik> fundamentally, I think all TX's should incur a cost.  that's the best way to deter spam, IMO.
585 2010-11-19 15:45:13 <m0mchil> your client detects you sending to yourself... it's trivial
586 2010-11-19 15:45:26 <m0mchil> but can't for others
587 2010-11-19 15:46:05 <edcba> tx fee should be dynamic
588 2010-11-19 15:46:16 <edcba> ie allow free tx
589 2010-11-19 15:46:31 <ArtForz> errr... tx fee is dynamic
590 2010-11-19 15:46:40 <edcba> and superior tx fee first
591 2010-11-19 15:46:51 <m0mchil> also, generators can't pose arbitrary fee... they decide whether to include or reject a TX based on the fee it provides
592 2010-11-19 15:46:55 <ArtForz> the current spammer just fills up the fee-less part of the block
593 2010-11-19 15:47:20 <edcba> ArtForz: the problem is bandwidth
594 2010-11-19 15:47:58 <ArtForz> currently... not really
595 2010-11-19 15:48:23 <ArtForz> but yeah, it could become a problem if someone bothers to optimize his spamming
596 2010-11-19 15:48:55 <ByteCoin> Problem definitely not bandwidth at the moment!
597 2010-11-19 15:49:35 <edcba> prioritize tx fees and limit propagation of free TX
598 2010-11-19 15:49:36 <jgarzik> spammer fills up fee-less part of the block, yes.  but doesn't the software also hold other spam TX's in memory, waiting for another block?
599 2010-11-19 15:49:42 <ArtForz> yes
600 2010-11-19 15:49:42 <edcba> that's all
601 2010-11-19 15:49:55 <ArtForz> btw, we should limit tx cache size
602 2010-11-19 15:49:57 <brocktice> Question, if everyone or most people mining start charging a 0.001BTC fee or whatever, will that help mitigate the issue?
603 2010-11-19 15:49:59 <jgarzik> ie. say I send out 1 million TX's, and don't care if they reach a block or not.  I'm using resources
604 2010-11-19 15:50:20 <ArtForz> otherwise a fast spam will cause a shitload of nodes to run out of memory
605 2010-11-19 15:50:25 <jgarzik> exactly
606 2010-11-19 15:50:37 <jgarzik> plus I'm forcing nodes to relay
607 2010-11-19 15:50:56 <ArtForz> btw, dont nodes also keep "orphan" transactions around?
608 2010-11-19 15:51:05 <edcba> that's why i said 1000 tx queue
609 2010-11-19 15:51:13 <m0mchil> ArtForz, do you plan making changes in your code about this?
610 2010-11-19 15:51:21 <edcba> and discard the rest
611 2010-11-19 15:51:27 <ArtForz> yeah
612 2010-11-19 15:51:29 <jgarzik> ArtForz: I think yes, periodically trying to get them into a block
613 2010-11-19 15:51:34 <m0mchil> or better wait for official solution?
614 2010-11-19 15:51:48 <ArtForz> and prioritise the tx cache similar to what we do for the miner now
615 2010-11-19 15:52:28 <ArtForz> = fee Tx first, then the rest prioritised by age of inputs and value
616 2010-11-19 15:52:43 <ArtForz> m0mchil: I'll wait for the official fix
617 2010-11-19 15:53:55 <ArtForz> hurrrm
618 2010-11-19 15:54:13 <ArtForz> at least we dont relay orphan TX
619 2010-11-19 15:54:50 <ArtForz> but the way it currently is, orphan TX cache means you can pretty much kill any node by IP if you have decent bandwidth
620 2010-11-19 15:55:28 <ArtForz> the txin/txout of a orphan TX dont have to refer to anything real
621 2010-11-19 15:56:04 <ArtForz> so a attacker can create completely bogus TX and flood a single node with em
622 2010-11-19 15:59:09 <MacRohard> they could just ping flood too
623 2010-11-19 15:59:12 <ArtForz> no
624 2010-11-19 15:59:19 <ArtForz> a ping flood just consumes bandwidth
625 2010-11-19 15:59:30 <ArtForz> a orphan tx flood consumes memory
626 2010-11-19 15:59:34 <MacRohard> hmm ok
627 2010-11-19 16:00:06 <ArtForz> with the current code if they just send enough of em long enough the victim node will run out of memory
628 2010-11-19 16:00:20 <jgarzik> ArtForz: will orphan TX's get relayed, I wonder?
629 2010-11-19 16:00:23 <ArtForz> no
630 2010-11-19 16:00:36 <MacRohard> i guess some kind of firewall or banlist might be an idea
631 2010-11-19 16:00:46 <ArtForz> thats why I said it's a targeted attack
632 2010-11-19 16:00:51 <MacRohard> like if someone feeds you too many bad TXs just cut them off
633 2010-11-19 16:01:09 <ArtForz> yeah, we'll have to implement rate limiting some day
634 2010-11-19 16:01:44 <brocktice> ArtForz: do you know whether Diablo-D3 ever figured out his stop-at-one-block problem?
635 2010-11-19 16:02:28 <ArtForz> nope
636 2010-11-19 16:02:35 <ArtForz> I think he's still working on it
637 2010-11-19 16:03:01 <brocktice> hm, ok, thanks
638 2010-11-19 16:09:25 <gavinandresen> Hey y'all, sorry I wasn't paying attention, just caught up on the txn flood discussion.
639 2010-11-19 16:12:15 <ByteCoin> It's cool.
640 2010-11-19 16:12:21 <gavinandresen> I still like the idea of requiring some proof-of-work to put a free transaction on the network.  If you don't want to pay a txn fee, then you gotta pay up with some useless CPU time.
641 2010-11-19 16:12:37 <ByteCoin> Good idea
642 2010-11-19 16:12:41 <ArtForz> thats actually a pretty good idea
643 2010-11-19 16:12:46 <ArtForz> hashcash for TX :P
644 2010-11-19 16:12:59 <gavinandresen> Spammers then need to decide:  should I waste CPU time spamming transactions that I COULD be using to try to generate?
645 2010-11-19 16:13:31 <ArtForz> *and* it should be pretty easy to implement
646 2010-11-19 16:13:38 <MacRohard> you're talking about a proof-of-work that's easier than the block generation proofofwork?
647 2010-11-19 16:13:42 <ArtForz> yep
648 2010-11-19 16:13:54 <gavinandresen> Nodes could sort by transaction proof-of-work (more proof == higher priority)....
649 2010-11-19 16:13:59 <MacRohard> hmm.. why not just adjust the block proof of works so that they're easier?
650 2010-11-19 16:14:04 <gavinandresen> And yeah, much easier proof of work than block generation
651 2010-11-19 16:14:05 <ByteCoin> Yeah. I presume the hash of the transaction could be calculated by the client
652 2010-11-19 16:14:06 <MacRohard> and happen more frequently
653 2010-11-19 16:14:21 <ArtForz> MacRohard: because that'll solve exactly nothing
654 2010-11-19 16:14:37 <MacRohard> ARtForz, hmmm
655 2010-11-19 16:14:58 <xelister_> gavinandresen: you think spam would slow down gpu miners noticibly?
656 2010-11-19 16:15:04 <ArtForz> no
657 2010-11-19 16:15:17 <ArtForz> at least not m0mchils and mine
658 2010-11-19 16:15:34 <ByteCoin> Different nodes could refuse to accept and relay transactions unless the transaction hash started with a certain number of zeros
659 2010-11-19 16:15:41 <ArtForz> yep
660 2010-11-19 16:15:50 <ByteCoin> The number of zeros required could be specified bythe  node
661 2010-11-19 16:15:50 <gavinandresen> yep
662 2010-11-19 16:15:53 <xelister_> what does it do again?  sends invalid transactions stuff to node so that bitcoind spends time calculating stuff with it?
663 2010-11-19 16:16:14 <MacRohard> xelister_, yeah
664 2010-11-19 16:16:30 <ArtForz> the transactions are valid
665 2010-11-19 16:16:49 <MacRohard> well yae. or just shitloads of valid transactions to yourself or connected parties
666 2010-11-19 16:16:54 <ArtForz> yep
667 2010-11-19 16:17:06 <ArtForz> the latest spams seem to be of the x->x kind
668 2010-11-19 16:17:26 <ByteCoin> Of course, you could sidestep the bulk of the problem by transitioning to a "balance sheet" method.....
669 2010-11-19 16:17:41 <ArtForz> yes, lets just rewrite everyfuckingthing
670 2010-11-19 16:17:41 <gavinandresen> (anybody else have the Laurie Anderson song "Let X == X" in their head right now?)
671 2010-11-19 16:17:47 <ByteCoin> Over time you might find that simpler....
672 2010-11-19 16:17:54 <gavinandresen> ArtForz:  I've been really tempted....
673 2010-11-19 16:18:15 <gavinandresen> (but not by balance sheets)
674 2010-11-19 16:18:31 <ByteCoin> The reason for not rewriting everything is that there's lots to be learnt from the current system still.
675 2010-11-19 16:19:17 <MacRohard> it might be cool if nodes could do net settlement and then peopel could arbitrage fees
676 2010-11-19 16:19:24 <ByteCoin> gavinandresen: Ok briefly what's your main objection to balance sheets?
677 2010-11-19 16:19:51 <gavinandresen> I haven't thought about them enough to have a main objection.
678 2010-11-19 16:20:03 <MacRohard> like a node that has generated a block could discount pending transactions where tehy partially cancel each other out.. maybe.
679 2010-11-19 16:20:13 <ByteCoin> Fair enough...
680 2010-11-19 16:20:49 <gavinandresen> (but if I did, my main objection would probably be that they're not enough-better than the current scheme to justify a switch.  Technology isn't Bitcoin's problem right now, trust and acceptance and marketing and usability are)
681 2010-11-19 16:20:54 <MacRohard> it would be nice for anonymity even if it did nothing for anything else
682 2010-11-19 16:21:07 <ByteCoin> Agreed.
683 2010-11-19 16:21:41 <ArtForz> hrrrm... proritising TX by their hash looks like it should be quite easy
684 2010-11-19 16:21:57 <gavinandresen> Satoshi has ignored all of my suggestions that nodes drop "bad" transactions...  I bet he has a good reason for it.
685 2010-11-19 16:22:09 <ArtForz> define "bad"
686 2010-11-19 16:22:14 <ByteCoin> Bad defined as malformed?
687 2010-11-19 16:22:21 <ByteCoin> Or bad as undesirable
688 2010-11-19 16:22:32 <ByteCoin> I can understand not censoring the latter
689 2010-11-19 16:22:37 <ArtForz> nonstandard script != bad
690 2010-11-19 16:22:52 <gavinandresen> Bad as in orphan, or is a 0.01-value transaction whos great-great-great grandparents are not yet part of any block....
691 2010-11-19 16:22:59 <ByteCoin> Yes Artforz otherwise it defeats the object of the script
692 2010-11-19 16:23:23 <gavinandresen> Bad as in "smells like spam"
693 2010-11-19 16:23:40 <ArtForz> the current orphan TX handling is pretty bad as it is
694 2010-11-19 16:23:46 <ByteCoin> gavin: Just to clarify, I presume you can't tell if the great-great grandparents are missing or just the parents...
695 2010-11-19 16:24:18 <ByteCoin> Otherwise I'm seriously misunderstanding something
696 2010-11-19 16:24:20 <gavinandresen> ByteCoin:  I'm talking about the case where you get several transactions between blocks, all of which are descendants....
697 2010-11-19 16:24:39 <gavinandresen> (the simple way of spamming:  just shave of 0.01 bitcoins from a larger coin, and send them all out...)
698 2010-11-19 16:24:46 <ByteCoin> Oh I see...
699 2010-11-19 16:24:47 <ArtForz> imo its a bit of a misfeature
700 2010-11-19 16:25:30 <ByteCoin> Requiring valid transactions to have a parent in a block would merely slow down the spam.
701 2010-11-19 16:25:37 <gavinandresen> yep
702 2010-11-19 16:25:52 <ByteCoin> You'd just have to divide a larger balance into small coins etc...
703 2010-11-19 16:26:03 <ArtForz> yes, but it would make orphan tx pretty much impossible
704 2010-11-19 16:26:38 <ByteCoin> Orphan tx can arise naturally if different transactions take different routes across the network
705 2010-11-19 16:26:46 <ArtForz> currently, yes
706 2010-11-19 16:26:49 <edcba> it slows down tx too..
707 2010-11-19 16:27:16 <ArtForz> yep
708 2010-11-19 16:27:17 <ByteCoin> You could avoid retransmitting orphan tx until you have the parent...
709 2010-11-19 16:27:28 <ByteCoin> Get's hard to test this stuff though...
710 2010-11-19 16:27:29 <ArtForz> no node transmits orphan TX
711 2010-11-19 16:27:43 <ArtForz> err... forwards
712 2010-11-19 16:27:50 <ByteCoin> Oh ok....
713 2010-11-19 16:29:23 <ArtForz> the original creator resends them periodically, hoping to reach a generating node that has the parent/grandparent/...
714 2010-11-19 16:29:35 <gavinandresen> So the question I'm thinking about is:  if transactions DID have proof-of-work (maybe just a nonce, and you do as much work as you like to make the transaction hash small)... would that fix anything?
715 2010-11-19 16:29:51 <ByteCoin> I think it would...
716 2010-11-19 16:30:06 <gavinandresen> Spammers could still spam zero-proof-of-work transactions.   If they were just ignored... would bad things happen?
717 2010-11-19 16:30:06 <Kiba> transaction as proof of work?
718 2010-11-19 16:30:20 <gavinandresen> I mean, ignored by SOME of the network nodes but not others.
719 2010-11-19 16:30:23 <ByteCoin> Don't give them enough room for a nonce. Make them keep resigning their txins!
720 2010-11-19 16:30:34 <ArtForz> they'd accumulate in the tx or orphan caches
721 2010-11-19 16:31:17 <ArtForz> so, limit those in size and prioritise by hash there, too
722 2010-11-19 16:31:24 <gavinandresen> So lets say one of them ends up in a block....  how do the nodes that were ignoring them get them?  I haven't looked at or thought about how trasnactions percolate across the net enough
723 2010-11-19 16:31:34 <ByteCoin> I'm thinking about adverse side effects.. It's complicated.
724 2010-11-19 16:31:54 <ArtForz> a Tx can only get in a block if all it's ancestors are also in that or previous blocks
725 2010-11-19 16:32:09 <ArtForz> = we can never have a orphan TX in a block
726 2010-11-19 16:32:24 <gavinandresen> YEah, these wouldn't be orphans, just transactions with low or no proof-of-work.
727 2010-11-19 16:32:32 <ArtForz> well, the block is broadcast as usual
728 2010-11-19 16:32:39 <ArtForz> why would a nod ignore a TX in a block?
729 2010-11-19 16:32:45 <ArtForz> that makes 0 sense
730 2010-11-19 16:32:49 <ByteCoin> yeah what artforz says
731 2010-11-19 16:33:06 <gavinandresen> ah, all transactions are rebroadcast when the block is announced?
732 2010-11-19 16:33:15 <gavinandresen> I thought it was just the block header and merkle root...
733 2010-11-19 16:33:30 <ArtForz> all tx that made it into that block are part of the block, and the whoel block is broadcast
734 2010-11-19 16:33:32 <gavinandresen> (like I said, I haven't look carefully at the net stuff...)
735 2010-11-19 16:33:46 <gavinandresen> cool.
736 2010-11-19 16:34:35 <ByteCoin> Anyway, if that were not the case and just the block header were broadcast then clients would have to inquire what the individual transactions were if they hadn't got them already
737 2010-11-19 16:35:09 <ByteCoin> Otherwise it'd be "Hmm... Here's a new block but we don't have a clue what's in it"
738 2010-11-19 16:35:16 <ArtForz> yep
739 2010-11-19 16:35:46 <ByteCoin> Artforz are you sure all the transactions in the block are transmitted with the block. It sounds sub-optimal
740 2010-11-19 16:35:50 <ArtForz> yes
741 2010-11-19 16:35:54 <gavinandresen> But this isn't gnutella -- there's no "hey network, find me transaction XYZ and return it to me, please" recursive-search-doohickey...
742 2010-11-19 16:36:38 <ArtForz> I wrote a python half-a-node thats capable of parsing tx and block messages, all tx contained in a block are included in the block message
743 2010-11-19 16:36:44 <ByteCoin> Artforz, the thing is that most nodes already know about all the block transactions and can match them up with stuff in their memory by the hashes.
744 2010-11-19 16:36:58 <ArtForz> erm, so?
745 2010-11-19 16:37:05 <ByteCoin> No need to send all the transactions again.
746 2010-11-19 16:37:16 <ArtForz> well, thats how it currently is
747 2010-11-19 16:37:23 <ByteCoin> Fair enough...
748 2010-11-19 16:38:08 <ArtForz> so thats not really a problem
749 2010-11-19 16:38:46 <ArtForz> adding a explicit nonce field to tx is imo a bad idea and breaks compatibility
750 2010-11-19 16:39:02 <gavinandresen> Next possible problem that comes to mind with proof-of-work transactions:  could somebody decide to be annoying and pre-compute a bunch of high-proof-of-work transactions, then flood the net with them all at once?
751 2010-11-19 16:39:10 <ArtForz> yes
752 2010-11-19 16:39:18 <ByteCoin> As just hinted. Explicit nonce not required.
753 2010-11-19 16:39:26 <ArtForz> and he'll have expended a significant amount of CPU time on it
754 2010-11-19 16:39:58 <ByteCoin> Gavinandresen: Yes possible.. Can't think of an obvious workaround
755 2010-11-19 16:40:04 <gavinandresen> If I recall, transactions don't include any notion of time; it'd be nice if old transactions were not allowed.
756 2010-11-19 16:40:17 <ByteCoin> Problem for offline clients..
757 2010-11-19 16:40:28 <ByteCoin> One way data transmission
758 2010-11-19 16:40:43 <gavinandresen> They'll have to go online SOME time to submit their transactions.  Proof of work could happen then.
759 2010-11-19 16:40:56 <gavinandresen> (or they pay a fee)
760 2010-11-19 16:41:08 <ByteCoin> They can send the transactuions without needing data back.
761 2010-11-19 16:41:26 <ByteCoin> If you introduce "time" they need to get data before sending tx
762 2010-11-19 16:41:50 <ArtForz> okay, so... offline nodes that dont know the current time?
763 2010-11-19 16:42:13 <gavinandresen> That seems unlikely.  EVERYTHING has a clock in it these days.
764 2010-11-19 16:42:14 <ByteCoin> I think that "the time" is not the issue as you can precompute the time in the future easily!
765 2010-11-19 16:42:23 <gavinandresen> ByteCoin:  good point!
766 2010-11-19 16:42:35 <ByteCoin> You'd need some data keyed to the latest block.
767 2010-11-19 16:42:39 <gavinandresen> Better would be... block number and hash
768 2010-11-19 16:42:48 <gavinandresen> (for a recent block)
769 2010-11-19 16:42:57 <ArtForz> imo adding block#s is a bad idea
770 2010-11-19 16:43:06 <ByteCoin> Do you REALLY want to go down this route gavin?
771 2010-11-19 16:43:16 <gavinandresen> No
772 2010-11-19 16:43:21 <gavinandresen> Just thinking out loud
773 2010-11-19 16:43:35 <ByteCoin> Cool. as long as you have not closed your mind
774 2010-11-19 16:43:37 <brocktice> hey, maybe you guys ignored me because it was a dumb idea earlier, but what if the many people in here who mine just started charging very small tx fees?
775 2010-11-19 16:43:46 <brocktice> I'm probably missing something though, right?
776 2010-11-19 16:43:51 <edcba> lol
777 2010-11-19 16:44:02 <ByteCoin> No you're not missing anything brocktice
778 2010-11-19 16:44:14 <brocktice> Would that help?
779 2010-11-19 16:44:17 <ArtForz> yes
780 2010-11-19 16:44:28 <ByteCoin> I think we all like the idea of discouraging things without actually having a "fine" for doing it.
781 2010-11-19 16:44:39 <ArtForz> and it still means we have ever-growing tx and orphan caches
782 2010-11-19 16:44:45 <brocktice> Ah right, there's still that
783 2010-11-19 16:44:49 <brocktice> I don't see any reason not to limit that.
784 2010-11-19 16:45:05 <ArtForz> just randomly dropping tx from the caches is imo a bad idea
785 2010-11-19 16:45:15 <edcba> why ?
786 2010-11-19 16:45:28 <brocktice> Is there a non-random way that makes sense and is not difficult to implement?
787 2010-11-19 16:45:39 <ArtForz> because of our current forwarding scheme
788 2010-11-19 16:45:45 <ByteCoin> We;'re thinking about it....
789 2010-11-19 16:46:03 <sec^nd> hello
790 2010-11-19 16:46:13 <brocktice> Also, do we have an IP or IPs that we can find where this is originating?
791 2010-11-19 16:46:28 <edcba> no
792 2010-11-19 16:46:36 <ByteCoin> To a certain extent it's not relevant
793 2010-11-19 16:46:50 <ByteCoin> What do we do with the info when we have it?
794 2010-11-19 16:46:55 <edcba> but you can do stats by node
795 2010-11-19 16:46:56 <brocktice> Well, there are grey hat methods for countering such people.
796 2010-11-19 16:46:59 <gavinandresen> I agree with ByteCoin
797 2010-11-19 16:47:20 <gavinandresen> The bad guys have no problem getting as may IPs as they want.
798 2010-11-19 16:47:32 <ByteCoin> The attack is just rubbing our noses in the fact that there's a problem
799 2010-11-19 16:47:38 <edcba> the more spam tx fees a node send the more you drop
800 2010-11-19 16:48:13 <edcba> so it won't propagate
801 2010-11-19 16:48:21 <ArtForz> how about this: re-enable the ability to replace a TX with one with identical in/outputs, limit cache sizes, prioritise tx in caches and mining by fee + tx hash "lowness" + age of tx inputs + value of tx
802 2010-11-19 16:49:14 <ArtForz> we need #1 so a node that sent a TX with a high hash can replace it with one with a lower hash if it wants to
803 2010-11-19 16:49:15 <ByteCoin> Thinking....
804 2010-11-19 16:49:18 <Diablo-D3> http://www.avtimer.com/images/Humor/Stealth%20F-22%20First%20Photo.jpg
805 2010-11-19 16:49:21 <Diablo-D3> ehehehh
806 2010-11-19 16:49:35 <Diablo-D3> ArtForz: so the java engine is now spitting out h == 0
807 2010-11-19 16:49:43 <ArtForz> Diablo-D3: cool
808 2010-11-19 16:49:59 <brocktice> Diablo-D3: clocks are bumped and I dropped in a 5770
809 2010-11-19 16:50:03 <gavinandresen> ArtForz: sounds good to me.  But network design isn't something I know anything about...
810 2010-11-19 16:50:06 <ByteCoin> I quite like the idea that hash quality could be used in place of a fee....
811 2010-11-19 16:50:09 <brocktice> currently drawing 1015W at the plug
812 2010-11-19 16:50:15 <Diablo-D3> ArtForz: so, 75 should get me a testnet block every 6 minutes?
813 2010-11-19 16:50:21 <brocktice> Diablo-D3: so whenever you've got a miner that'll find more than one block I can fire it up
814 2010-11-19 16:50:22 <ArtForz> I'm still trying to figure out if dropping tx from caches would cause a problem
815 2010-11-19 16:51:00 <ArtForz> but I dont really think so
816 2010-11-19 16:51:12 <ArtForz> restarting a node has roughly the same effect
817 2010-11-19 16:51:18 <gavinandresen> ArtForz:  I can't see why it would be a problem.
818 2010-11-19 16:51:41 <edcba> as long origin/dest keep them  it's ok
819 2010-11-19 16:51:44 <brocktice> I just had a brilliant idea
820 2010-11-19 16:51:47 <Diablo-D3> brocktice: I still gotta test that
821 2010-11-19 16:52:01 <brocktice> My miner puts out a lot of hot air, warming the room, but it moves upward and my toes are still cold.
822 2010-11-19 16:52:09 <gavinandresen> I'm away for lunch for a while....
823 2010-11-19 16:52:11 <brocktice> So I'll route some ducts from the back end and have them blow out at my feet
824 2010-11-19 16:52:18 <brocktice> = footwarmer
825 2010-11-19 16:52:25 <Diablo-D3> brocktice: ....
826 2010-11-19 16:52:34 <ByteCoin> gavin: are you going to make a forum post about this? or should I?
827 2010-11-19 16:52:36 <Diablo-D3> Attempt 1:58:30 AM 252 14 232 169 240 211 120 99 124 212 174 94 170 156 20 125 170 200 251 63 223 38 180 202 137 173 37 250 0 0 0 0
828 2010-11-19 16:52:37 <Diablo-D3> so, fuck yeah
829 2010-11-19 16:52:46 <brocktice> Diablo-D3: What, it's a great idea
830 2010-11-19 16:52:53 <Diablo-D3> ArtForz: btw, you know what I screwed up?
831 2010-11-19 16:53:00 <Diablo-D3> ArtForz: I sent all 128 bytes of that 80 byte header.
832 2010-11-19 16:53:07 <ByteCoin> Diablo: Tell us.
833 2010-11-19 16:53:14 <ArtForz> oh. oops.
834 2010-11-19 16:53:21 <ByteCoin> You foind that out ages ago
835 2010-11-19 16:53:28 <Diablo-D3> ByteCoin: I found that out yesterday you moron
836 2010-11-19 16:53:31 <ArtForz> brocktice: next episode "how to heat your house with $10k of GPUs"
837 2010-11-19 16:53:38 <Diablo-D3> art wasnt here to hear it
838 2010-11-19 16:53:46 <brocktice> ArtForz: That'd be great, post it on youtube
839 2010-11-19 16:53:49 <ByteCoin> yesterday= ages.
840 2010-11-19 16:54:11 <Diablo-D3> ArtForz: so I should find one roughly every 6 minutes?
841 2010-11-19 16:54:49 <ArtForz> Diablo-D3: secperblock = difficulty * 2**32 / hashpersec
842 2010-11-19 16:54:58 <ArtForz> testnet is currently diff 5.33
843 2010-11-19 16:55:53 <ArtForz> so 305 sec at 75Mh/s
844 2010-11-19 16:56:30 <ArtForz> = about 5 min/block
845 2010-11-19 16:57:28 <ArtForz> grat
846 2010-11-19 16:58:14 <ArtForz> the spammer now does a->b->a
847 2010-11-19 16:58:34 <ByteCoin> I wonder why they changed....
848 2010-11-19 16:58:36 <brocktice> Heh, I wonder if they're lurcking
849 2010-11-19 16:58:38 <brocktice> *lurking
850 2010-11-19 16:58:45 <brocktice> occurred to me before
851 2010-11-19 16:58:49 <ArtForz> yeah
852 2010-11-19 16:59:01 <ArtForz> hmmmm
853 2010-11-19 16:59:22 <ArtForz> actually it should be pretty trivial to locate the spammers IP using my python not-quite-a-node
854 2010-11-19 16:59:40 <edcba> indeed
855 2010-11-19 16:59:40 <Kiba> transaction fee as minicule as 0.0001 BTC will eventually cause them to run out of money
856 2010-11-19 16:59:43 <ByteCoin> not seeing it Artforz : I'm seeing a->a->a and b->b->b
857 2010-11-19 17:00:06 <ArtForz> whoops, you're right
858 2010-11-19 17:00:13 <ArtForz> nodes delay forwards by about 0.2s
859 2010-11-19 17:01:09 <ByteCoin> They started using 0.01 instead of 0.06 last night. I presume they're running multiple computers to stop the client slowing down
860 2010-11-19 17:01:15 <ArtForz> yeah
861 2010-11-19 17:01:23 <ArtForz> they lagged the hell out of their client :P
862 2010-11-19 17:01:38 <ByteCoin> Thank god for super slow code
863 2010-11-19 17:01:47 <theymos> They could just send their balance to a new client to get a fresh wallet.
864 2010-11-19 17:02:03 <ByteCoin> I think that's what they must do periodically.
865 2010-11-19 17:02:11 <ArtForz> yep
866 2010-11-19 17:02:30 <ByteCoin> You realize that technically there's no barrier to sending out millions of valid transactions a second
867 2010-11-19 17:02:31 <Kiba> somebody stresstesting the network?
868 2010-11-19 17:02:38 <ByteCoin> It's not CPU bound
869 2010-11-19 17:02:53 <brocktice> It must be at some point
870 2010-11-19 17:02:57 <ArtForz> oh, it is
871 2010-11-19 17:02:59 <brocktice> but that point is probably very, very high
872 2010-11-19 17:03:07 <ByteCoin> Yes kiba. I think all the blocks are full to 50k
873 2010-11-19 17:03:11 <ArtForz> ECDSA signing isn't exactly very fast
874 2010-11-19 17:03:32 <ByteCoin> Depends what you mean by fast
875 2010-11-19 17:03:43 <ArtForz> > a few kTX/s
876 2010-11-19 17:03:44 <ByteCoin> Not as fast as a hash
877 2010-11-19 17:04:44 <ByteCoin> I won't commit to it but I think it's probably in the hundreds of kTX/s if you want
878 2010-11-19 17:04:59 <Kiba> price is on...the rise
879 2010-11-19 17:05:03 <ByteCoin> Especially if you cyt some corners...
880 2010-11-19 17:05:19 <Kiba> btcex have extremely low volume
881 2010-11-19 17:05:25 <Kiba> must be not many Russians
882 2010-11-19 17:05:32 <Kiba> or maybe their money is worthless
883 2010-11-19 17:05:40 <ArtForz> well, all it does is consume bandwidth and disk space
884 2010-11-19 17:05:52 <theymos> ArtForz: Could you drop all transactions based on coins from the spammer? That'd clear up some blocks for real use.
885 2010-11-19 17:06:27 <ArtForz> kinda pointless
886 2010-11-19 17:06:38 <ByteCoin> Artforz; I've been poring over wireshark dumps to try to make a compatible client. Is your half client code available?
887 2010-11-19 17:06:44 <ArtForz> all that'd accomplish is 1-in-5 blocks nearly empty
888 2010-11-19 17:06:52 <ArtForz> ByteCoin: sec
889 2010-11-19 17:06:55 <Kiba> is it possible that we move this channel to #bitcoin?
890 2010-11-19 17:06:58 <ByteCoin> cheeres!
891 2010-11-19 17:07:29 <Kiba> I heard that we don't need IRC to connect to the bitcoin network anymore
892 2010-11-19 17:07:48 <ByteCoin> Still think it's heavily used though...
893 2010-11-19 17:08:48 <theymos> Kiba: Bitcoin hasn't used that channel since (long) before 0.3.10. It uses a channel on irc.lfnet.org now.
894 2010-11-19 17:13:46 <ByteCoin> I could see from wireshark that there was plenty of IRC activity. It seems like #bitcoin on this server is quite quiet though.
895 2010-11-19 17:16:36 <ByteCoin> Artforz: From wireshark I was getting the impression that there might be some important variation in network data format based on client version. Is that what you've seen?
896 2010-11-19 17:17:39 <ArtForz> yes
897 2010-11-19 17:17:40 <ArtForz> sec
898 2010-11-19 17:19:37 <Kiba> I see random username
899 2010-11-19 17:19:40 <Kiba> must be old clients
900 2010-11-19 17:20:37 <theymos> Kiba: The usernames are encoded IP addresses.
901 2010-11-19 17:20:41 <ArtForz> http://pastebin.com/ZSM7iHZw
902 2010-11-19 17:21:15 <ByteCoin> Thx. Looking at it now
903 2010-11-19 17:21:21 <ArtForz> all it does currently is connect to a node and receive TX and blocks as they come by
904 2010-11-19 17:22:40 <ByteCoin> Oooh. There's quite a bit of it!
905 2010-11-19 17:23:03 <ArtForz> yeah, it can parse/build all curerently used network messages
906 2010-11-19 17:23:16 <ArtForz> I'm just not using most of it in this example
907 2010-11-19 17:24:27 <brocktice> I was thinking of building a python client
908 2010-11-19 17:24:39 <brocktice> Maybe with PyQT
909 2010-11-19 17:24:52 <ArtForz> I have another one that actually has the script engine, keeps a block chain, verifies transactions, has a tx cache, ...
910 2010-11-19 17:25:49 <ByteCoin> Whoa! That's a lot
911 2010-11-19 17:26:00 <ArtForz> and yes, the code is about as horrible as the rest
912 2010-11-19 17:26:12 <ArtForz> I wrote most of this as I was reverse engineering the protocol
913 2010-11-19 17:26:28 <ByteCoin> Code doesn't look bad.
914 2010-11-19 17:27:34 <brocktice> Diablo-D3: How's it going on testnet?
915 2010-11-19 17:28:03 <Diablo-D3> not sure
916 2010-11-19 17:28:17 <Diablo-D3> though Ive discovered a small problem
917 2010-11-19 17:28:52 <Diablo-D3> http://pastebin.com/jvCBrRAk
918 2010-11-19 17:29:18 <Diablo-D3> and I suspect those three valid attempts is one for each executor
919 2010-11-19 17:29:26 <ArtForz> "whoops"
920 2010-11-19 17:29:45 <Diablo-D3> not "whoops" as much as "wtf"
921 2010-11-19 17:29:52 <Diablo-D3> Im not sure what Im looking at
922 2010-11-19 17:30:55 <jgarzik> Kiba: IRC is used for network address seeding.  It's never been a requirement, but absent the IPv4 addresses compiled into the bitcoin binary, or manually specifying addresses on the command line, IRC is all you have for bootstrapping.
923 2010-11-19 17:33:36 <jgarzik> theymos: does BBE show block size anywhere?
924 2010-11-19 17:33:42 <theymos> jgarzik: No.
925 2010-11-19 17:34:04 <nanotube> hmm... well at least, it's a good thing we are forced to deal with the tx spam problem now.
926 2010-11-19 17:34:29 <ByteCoin> I would have been happier waiting
927 2010-11-19 17:34:38 <nanotube> theymos: btw, i notice you have not included the tx spam problem in the weaknesses pages :)
928 2010-11-19 17:34:41 <nanotube> page
929 2010-11-19 17:35:31 <theymos> nanotube: It's not really a problem. No one is being prevented from sending transactions.
930 2010-11-19 17:36:00 <nanotube> theymos: well, it would be a problem if they spam up to the max block size.
931 2010-11-19 17:36:12 <theymos> They'd have to pay fees before that.
932 2010-11-19 17:36:18 <nanotube> theymos: artforz doesn't require fees
933 2010-11-19 17:36:21 <nanotube> so they can spam up to the max
934 2010-11-19 17:36:25 <ArtForz> nope
935 2010-11-19 17:36:50 <ArtForz> I only dont require the 0.01 fee for < 0.01 output TX
936 2010-11-19 17:36:53 <nanotube> and otherwise, they can just spam up to the no-fee limit, and then force everyone /else/ to pay the fees.
937 2010-11-19 17:36:57 <ArtForz> and I think I can stop doing that now
938 2010-11-19 17:36:58 <nanotube> ArtForz: ah ok
939 2010-11-19 17:37:05 <Diablo-D3> ArtForz: any ideas?
940 2010-11-19 17:37:19 <ArtForz> Diablo-D3: for what?
941 2010-11-19 17:37:25 <nanotube> and yes, i think you can, since the viral-nature of the minitx has been fixed in .13
942 2010-11-19 17:37:26 <Diablo-D3> why this has its head up its ass
943 2010-11-19 17:37:44 <ArtForz> yep
944 2010-11-19 17:38:23 <jgarzik> is 219 transactions per block the maximum one can achieve without paying tx fees?
945 2010-11-19 17:38:48 <ByteCoin> Looks like it. Makes sense at around 220 bytes per transaction
946 2010-11-19 17:40:06 <jgarzik> blacklist abusive coins :)
947 2010-11-19 17:40:13 <Diablo-D3> jgarzik: not particularly a good idea
948 2010-11-19 17:40:22 <Diablo-D3> everyones should just up their transaction costs
949 2010-11-19 17:40:29 <ByteCoin> Diablo: Agreed
950 2010-11-19 17:40:43 <ByteCoin> Not agreed about increasing tx costs
951 2010-11-19 17:40:43 <theymos> jgarzik: Then we'd have to ban everyone who uses the system illegally, probably.
952 2010-11-19 17:40:53 <brocktice> Diablo-D3: that's what I said, at least in the short term
953 2010-11-19 17:41:32 <Diablo-D3> it'd take away their coin.
954 2010-11-19 17:41:47 <brocktice> They may not have any to begin with
955 2010-11-19 17:41:50 <brocktice> then it would just stop them
956 2010-11-19 17:42:15 <jgarzik> theymos: that outcome never arose from standard anti-spam fighting...  it tends to be network operations and engineers who wind up fighting this sort of network abuse.
957 2010-11-19 17:42:27 <Diablo-D3> brocktice: how are they continually sending a coin to themselves without one?
958 2010-11-19 17:42:35 <brocktice> Diablo-D3: let me test
959 2010-11-19 17:42:49 <brocktice> oh yeah
960 2010-11-19 17:42:57 <brocktice> ok so they have at least some fraction of a bitcoin